Project summary/abstract Essential genes are fundamental to genetics and functional genomics. Systematic knockout studies in yeast defined the first complete set of genes essential for cellular proliferation, and subsequent surveys of how gene essentiality varied across environmental and genetic backgrounds revealed foundational principles of functional genomics: that ?synthetic lethality? arises when one gene becomes essential in the presence of another gene's mutation or loss of function, and that genes operating in the same biological processes tend to have the same loss-of-function phenotypes when assayed across diverse backgrounds. The adaptation of the CRISPR/Cas9 system to humans has rendered our genome tractable, and in my postdoctoral training and in my current position as Assistant Professor at MD Anderson Cancer Center, I have made fundamental contributions advances in CRISPR screening. I led the first gene knockout study to identify both core and context-specific essential genes in cancer cells (Hart et al., Cell, 2015), and led the informatics effort that identified FZD5 as a specific vulnerability in RNF43-mutant pancreatic cancer (Steinhart et al., Nat Med, 2017). I designed all CRISPR reagents used in these studies, and subsequently integrated empirical data across many published screens to create a much smaller, vastly more efficient library (TKOv3; available on Addgene). My lab has advanced the state of the art in CRISPR informatics by developing algorithms to classify essential genes and to identify drug-gene interactions, and we have defined benchmarks of gold-standard essential and nonessential genes that have been adopted by every major screening study. The CRISPR screening effort in human cells is beginning to bear fruit, with high-quality data available from hundreds of cell lines. We seek to apply our combined expertise in integrative analysis and high- throughput biology to explore questions about the variation in gene essentiality across cellular lineage, genotype, and environment. As with yeast, groups of genes with similar knockout fitness profiles are likely involved in the same biological processes, providing an avenue for deciphering gene function. One-third of all protein-coding genes are constitutively and invariantly expressed, yet half of these show no knockout phenotype. Many are likely buffered by paralogs, potentially a rich source of synthetic lethal interactions. Core essentials, required in every cell, are more sensitive to perturbation when hemizygously deleted in cancer cells, which may help explain from first principles the fitness constraints on copy number rearrangement in cancer cells. Globally, patterns of shared genetic vulnerability are likely to reveal unexpected tumor subtypes, a key goal of our data-driven, network-based integrative analytical approach. Finally, we seek a predictive, process-level model of gene essentiality that can explain variations across lineage and genotype, and that further can be used to develop reduced-representation CRISPR reagents that enable high-information, low- cost screening approaches for more focused biological applications.